Sediments of Racism in Public Discourse: Operationalization, Empirical Assessment of its Occurrence and Effects
Research question/goal:
In political communication research, racial discrimination in public discourse is receiving in-creasing attention. Yet, existing quantitative research often avoids focusing on the concept of racism directly and measures only specific subdimensions of it (e.g., the stereotype content model). One explanation for this is that racism occurs as traces of a fragmented ideology in mediated discourse, with its different components hardly ever present within one individual mediated message. Based on a framework of 11 dimensions, the project develops an operationalisation of racism in public discourse that covers all theoretically meaningful subdimensions of the concept using automated content analysis with adapter-based transformer models. A large, heterogeneous news corpus is used to benchmark these measures and give an empirical assessment of the occurrence of racism in the German public sphere. The project then maps different areas of the German mediated public sphere, analysing in which combinations the message features under study occur in them. Subsequently, experimental research is employed to test a media effects model that accounts for the fragmented nature of racism in media messages by drawing on priming and schema theory.
Current stage:
The project proposal was still under review by the DFG (German Research Foundation) at the end of 2025, and a funding decision is not expected before spring 2026. In the meantime, the project team is focusing on the development of new methodological routines for automated content analysis using large language models (LLMs), which will be used within the project, and of the corresponding infrastructure. Among other things, an R package ('bacchuss') was created in 2025 and made publicly available via GitHub, and a server for the local use of LLMs for classification tasks was set up with the support of the university’s IT department.